Mortality Not Correlated With
Paclitaxel Exposure
An Independent Patient-Level Meta-Analysis of a
Drug-Coated Balloon
Peter A. Schneider, MD,aJohn R. Laird, MD,bGheorghe Doros, P
HD,c,dQi Gao, MS,cGary Ansel, MD,e
Marianne Brodmann, MD,fAntonio Micari, MD, PHD,gMehdi H. Shishehbor, DO, MPH, PHD,hGunnar Tepe, MD,i
Thomas Zeller, MD, PHDj
ABSTRACT
BACKGROUNDFive years of prospective clinical trials confirm that the paclitaxel drug-coated balloon (DCB) (IN.PACT Admiral, Medtronic, Dublin, Ireland) is safe and effective to treat femoropopliteal artery disease. A recent meta-analysis of heterogeneous trials of paclitaxel-based balloons and stents reported that they are associated with increased mortality and that higher doses are linked to higher mortality from 2 to 5 years.
OBJECTIVESThe purpose of this study was to determine if there is a correlation between paclitaxel exposure and mortality by conducting an independent patient-level meta-analysis of 1,980 patients with up to 5-year follow-up.
METHODSData from 2 single-arm and 2 randomized independently adjudicated prospective studies of a paclitaxel DCB (n¼ 1,837) and uncoated percutaneous transluminal angioplasty (PTA) (n ¼ 143) were included. Analyses of baseline, procedure, and follow-up data of individual patients were performed to explore correlations of paclitaxel dose with long-term mortality. Survival time by paclitaxel dose tercile was analyzed with adjustment of inverse probability weighting to correct baseline imbalances and study as random effect. A standard cohort was defined to compare DCB- and PTA-treated patients with similar characteristics by applying criteria from pivotal studies (n¼ 712 DCB, n ¼ 143 PTA).
RESULTSA survival analysis stratified nominal paclitaxel dose by low, mid, and upper terciles; mean doses were 5,019.0, 10,007.5, and 19,978.2mg, respectively. Rates of freedom from all-cause mortality between the 3 groups through 5 years were 85.8%, 84.2%, and 88.2%, respectively (p¼ 0.731). There was no significant difference in all-cause mortality between DCB and PTA through 5 years comparing all patients (unadjusted p¼ 0.092) or patients with similar characteristics (adjusted p¼ 0.188).
CONCLUSIONSThis independent patient-level meta-analysis demonstrates that this paclitaxel DCB is safe. Within DCB patients, there was no correlation between level of paclitaxel exposure and mortality. (Randomized Trial of IN.PACT Admiral Drug Coated Balloon vs Standard PTA for the Treatment of SFA and Proximal Popliteal Arterial Disease [INPACT SFA I],NCT01175850; IN.PACT Admiral Drug-Coated Balloon vs. Standard Balloon Angioplasty for the Treatment of Superficial Femoral Artery [SFA] and Proximal Popliteal Artery [PPA] [INPACT SFA II],NCT01566461; MDT-2113 Drug-Eluting Balloon vs. Standard PTA for the Treatment of Atherosclerotic Lesions in the Superficial Femoral Artery and/or Proximal Popliteal Artery [MDT-2113 SFA],NCT01947478; The IN.PACT SFA Clinical Study for the Treatment of Atherosclerotic Lesions in the Superficial Femoral Artery and/or Proximal Popliteal Artery Using the IN.PACT Admiral Drug-Eluting Balloon in a Chinese Patient Population,NCT02118532; and IN.PACT Global Clinical Study,
NCT01609296) (J Am Coll Cardiol 2019;73:2550–63) © 2019 The Authors. Published by Elsevier on behalf of the American College of Cardiology Foundation. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).
ISSN 0735-1097 https://doi.org/10.1016/j.jacc.2019.01.013
From theaDivision of Vascular and Endovascular Surgery, University of California at San Francisco, San Francisco, California; bAdventist Heart and Vascular Institute, St. Helena, California;cBaim Institute for Clinical Research, Boston, Massachusetts; dSchool of Public Health, Boston University, Boston, Massachusetts;eOhio Health, Riverside Methodist Hospital, Columbus, Ohio; fDivision of Angiology, Medical University Graz, Graz, Austria;gHumanitas Gavazzeni Hospital, Bergamo, Italy;hHeart & Vascular Institute, University Hospitals Cleveland Medical Center, and Case Western Reserve University School of Medicine, Cleveland, Ohio;iRoMed Clinic, Rosenheim, Germany; and thejUniversitäts-Herzzentrum Freiburg
–Bad Krozingen, Bad Krozingen, Ger-many. Data was transferred for independent analysis to the Baim Institute for Clinical Research, formerly HCRI. This analysis was
Listen to this manuscript’s audio summary by Editor-in-Chief Dr. Valentin Fuster on JACC.org.
A
potential longer-term safety concern has been raised about the use of paclitaxel-containing devices in the treatment of femo-ropopliteal artery occlusive disease in ameta-analysis of summary-level data (1). A higher
inci-dence of late mortality was reported across studies comparing paclitaxel drug-coated balloon (DCB) and drug-eluting stent (DES) therapies with un-coated percutaneous transluminal angioplasty (PTA) or bare-metal stent (BMS) controls. This difference in mortality was ascribed directly to paclitaxel. In
addition, the authors reported that“the risk of death
beyond one year also seemed to vary among
different paclitaxel dosages, being significantly
higher in the 3.5
m
g/mm2 devices compared withthe lower-dose devices”(1).
Prior to the publication of this meta-analysis, the body of evidence suggested that DCBs are safe and maintain a positive clinical benefit in the treatment of femoropopliteal disease in patients with peripheral arterial disease (PAD) and Rutherford 2 to 4 symp-toms, as reported in the SCAI Consensus Guidelines for Device Selection in Femoral-popliteal Arterial
In-terventions(2)and the ACC/AHA/SCAI/SIR/SVM 2018
Appropriate Use Criteria(3). Mortality rates across the individual trials of DCBs are well within the ranges reported in other adjudicated nondrug vascular de-vice trials and also reported in epidemiological studies of patients with PAD. However, results of the IN.PACT SFA investigational device exemption trial showed a transitorily higher frequency of mortality
between the DCB and PTA arms at 2 and 3 years(4,5).
These results were published, noting that this difference was no longer statistically
signifi-cant at 4 and 5 years (4–7). No higher
fre-quency of mortality was seen with DCBs at any time point in IN.PACT SFA Japan. All
major adverse events and deaths were
reviewed and adjudicated by an independent clinical events committee (CEC) and none were deemed related to the device.
As the recently published meta-analysis that suggested a safety concern was disad-vantaged by its derivation from published or pre-sented results of heterogeneous trials and devices, an independent assessment of adjudicated, individual
patient-level data is essential to determine if
paclitaxel-coated balloons are associated with an increased risk of mortality.
Reported herein is an independent patient-level analysis of patients treated with a paclitaxel DCB to investigate whether there is a correlation between paclitaxel exposure and mortality, and furthermore, whether increased paclitaxel dose is correlated with increased mortality.
METHODS
The Baim Institute for Clinical Research, Boston,
Massachusetts, independently performed all
analyses.
POOLED ANALYSIS STUDY DESIGN. This
indepen-dent individual patient-level meta-analysis was
designed to establish whether there is a correlation between nominal paclitaxel exposure and mortality
SEE PAGE 2564
A B B R E V I A T I O N S A N D A C R O N Y M S
CEC= clinical events committee
DCB= drug-coated balloon
MAE= major adverse event
PAD= peripheral artery disease
PTA= percutaneous transluminal angioplasty
funded by Medtronic. Dr. Schneider has served as a member of the advisory board for Medtronic, Abbott, and Boston Scientific; has served as a consultant for Surmodics, Silk Road Medical, Medtronic, Cardinal, CSI, and Profusa; and is a Chief Medical Officer for Intact Vascular and Cagent. Dr. Laird has served on the advisory board for Abbott Vascular and Boston Scientific; has served as a consultant for Abbott Vascular, Philips, Bard/Becton Dickinson, Boston Scientific, and Medtronic; and has received speaking fees from Medtronic. Dr. Doros is a full-time employee of the Baim Institute for Clinical Research; has served as a consultant for Pfizer and Supernus; and has served on the data safety monitoring board for Takeda, Alnylam, and NeoSync. Dr. Gao is a full-time employee of the Baim Institute for Clinical Research. Dr. Ansel has provided advisory and consultation services for Medtronic, CR Bard, Boston Scientific, Cook Medical, Philips Medical, WL Gore, Veryan Medical, Reflow Medical, Shockwave, Intact Vascular, Abbott Vascular, Alucent, Contego Medical, Cardiovascular Systems. Surmodics, Vascular Dynamics, Vatrix, and VIVA Physicians. Dr. Brodmann has received speaking honoraria from Bard Peripheral Vascular, Biotronik, Medtronic, Philips-Spectranetics, Shockwave, Bayer Healthcare, and VIVA Physicians; and is a consultant for Bard Peripheral Vascular, Biotronik, Medtronic, Shockwave, Intact Vascular, Bayer, Sanofi, and Philips-Spectranetics. Dr. Micari has served as a member of the Advisory Board for Medtronic and Boston Scientific; and has served as a consultant for Boston Scientific, Bard, and Terumo. Dr. Shishehbor has served as a consultant and a member of advisory board for Medtronic, Abbott Vascular, Boston Scientific, Philips, and Terumo. Dr. Tepe has received research grants from Medtronic, Bard, Bayer, Gore, Biotronik, and Boston Scientific; and has served as a compensated advisory board member for Medtronic and B Braun. Dr. Zeller has received honoraria from Abbott Vascular, Veryan, Biotronik, Boston Scientific, Cook Medical, Gore and Associates, Medtronic, Philips-Spectranetics, TriReme, and Shockwave; has served as a consultant for Boston Scientific, Cook Medical, Gore and Associates, Medtronic, Spectranetics, Veryan, Intact Vascular, B. Braun, Shockwave, Bayer, and Vesper Medical; has received research, clinical trial, or drug study funds from 480 biomedical, Bard Peripheral Vascular, Veryan, Biotronik, Cook Medical, Gore and Associates, Medtronic, Philips, Terumo, TriReme, Shock-wave, Med Alliance, Intact Vascular, and B. Braun; and owns common stock in Veryan and QT Medical.
in patients treated with the IN.PACT Admiral
pacli-taxel drug-coated balloon (Medtronic, Dublin,
Ireland) for the treatment of symptomatic femo-ropopliteal PAD. A total of 1,980 patients among heterogeneous ethnic populations comprised of 2 randomized controlled trials (RCTs) and 2 prospective single-arm studies conducted at 147 sites across 6
continents and 28 countries were included (Online
Table 1). The longest available follow-up data from 1,837 patients treated with this paclitaxel DCB and 143 patients treated with uncoated PTA were aggregated. IN.PACT SFA and MDT-2113 SFA Japan (IN.PACT SFA Japan) were prospective, multicenter, randomized,
single-blind trials comparing DCB with PTA (8,9).
IN.PACT SFA China was a prospective, multicenter,
pre-market single-arm study(10). IN.PACT Global was
a real-world, prospective, multicenter, single-arm
study comprised of pre-specified imaging cohorts
and a nonimaging clinical cohort (11). Patients
enrolled in IN.PACT Global were more likely to have long lesions, chronic total occlusions, and in-stent restenosis. In addition, greater numbers of patients with Rutherford Clinical Category (RCC) 4 and 5 were
part of IN.PACT Global, demonstrating the
nonequivalence of the DCB patient population
compared with the PTA population from the
pre-market RCTs (11). A notable difference among
studies is adherence to antiplatelet therapies; rec-ommendations were pre-specified in each group ac-cording to the current standard of practice. Aspirin was required indefinitely for IN.PACT SFA and
IN.PACT Global. Clopidogrel was required for
3 months in patients who received stents and for 1 month in patients who did not receive a stent in IN.PACT SFA, IN.PACT Global, and IN.PACT SFA Japan.
All major adverse events (MAEs), deaths, target limb amputations, reinterventions, and target lesion thromboses that occurred throughout each study follow-up period were independently adjudicated by a CEC representing varied specialists with no conflicts of interest (Syntactx, New York, New York; Baim/ HCRI, Boston, Massachusetts). Per the CEC’s manual of operations, device-related death was defined as a complication associated with the device design as it relates to placement, efficacy, or durability that may involve the delivery system.
These 4 studies comprise the largest, most
com-plete, and fully adjudicated database of
longitudinally-followed DCB patients to date. This provides added statistical power to evaluate whether paclitaxel played a role in mortality and whether paclitaxel dose had an effect on safety, although there is patient and lesion heterogeneity across
studies. Additionally, PTA patients were only
included in the 2 randomized trials in a 2:1 ratio. This
is appropriate powering for a 1-year patency
endpoint, and is appreciable for this analysis, but the low number of PTA patients may limit the power for any comparisons with the PTA group.
ENDPOINT DEFINITIONS. Study endpoint definitions were consistent across all IN.PACT Admiral DCB studies. All were assessed through the longest-available follow-up: 5 years for IN.PACT SFA, 3 years for IN.PACT SFA Japan, 3 years for IN.PACT Global, and 1 year for IN.PACT China. Follow-up of IN.PACT Global subjects to 5 years is in process. Assessment by Kaplan-Meier estimate included MAE and its compo-nent events: death from any cause, clinically-driven target vessel revascularization, target limb major amputation, and thrombosis. Clinically-driven rein-terventions were defined as reintervention at the target vessel and/or lesion because of symptoms or a
decrease in ankle-brachial index (ABI) by $20% or
>0.15 when compared with post-procedure baseline ABI. Previously, site-reported causes of mortality were reported and were not categorized by system class. The recent meta-analysis used the site-reported
causes of mortality and grouped them (1). For the
completeness of this analysis, CEC-adjudicated nar-ratives were reviewed individually, and these results were reported rather than the site-reported causes of death. Using the CEC narratives and the etiology of deaths provided, the causes of death were catego-rized by organ system classification.
STATISTICAL ANALYSES. All descriptive summaries were based on nonmissing assessments. Unless
otherwise specified, all baseline demographics and
clinical characteristics were summarized on a patient basis; lesion characteristics were summarized on a lesion basis. For baseline characteristics, continuous
variables were described as mean SD and were
compared by Student’s t-test or Wilcoxon rank sum test; dichotomous and categorical variables were described as counts and proportions and were compared by Fisher exact test or Cochran-Mantel-Haenszel test with modified ridit scores. Outcome analyses were performed at the patient level. The level of statistical significance was set at p < 0.05. Statistical analyses were performed using SAS version 9.4 (SAS Institute, Cary, North Carolina).
P a c l i t a x e l d o s e a n a l y s e s .Nominal paclitaxel dose
per balloon was defined according to the product
length and diameter as described in the IN.PACT
Admiral Instructions for Use(12). The balloon lengths
and diameters received by each patient were
paclitaxel dose per balloon was added together to define the total dose of paclitaxel received per patient in each index procedure.
To assess the association of the paclitaxel nominal dose received by each patient during the index pro-cedure and mortality over time, the group was segmented into 3 terciles based on the amount of paclitaxel received. The unadjusted survival in these 3 groups was estimated using the Kaplan-Meier method. Survival was further compared among dose terciles with the use of propensity scores and
inverse-probability-of-treatment-weighting (IPTW)
adjust-ment. IPTW was used rather than adjustment with multiple covariates, as adjustment could potentially lead to lack of model convergence when a small number of events are observed. Details on the
pro-pensity score model are included in the Online
Appendix. Frailty models, which are extensions of the proportional hazards model, were used to account for clustering of patients by study.
M u l t i v a r i a b l e a n a l y s i s .To identify predictive fac-tors for death in all DCB-treated patients,
multivari-able Cox regression models with frailty were
performed. Baseline variables with>15% missing data
were excluded from the analysis and included vari-ables with missing data were imputed: simple impu-tation was applied for missing values, stratified by study, sex, and treatment arm, using the mean for continuous variables or most common values for dichotomous or categorical variables of the non-missing values. Multivariate predictors started from the exact same set of variables used to derive IPTW between paclitaxel dose tercile and were chosen by eliminating 1 of the covariates with the largest p value at a time, until all covariates left in the model except for paclitaxel dose tercile have p value<0.20. Hazard ratios (HRs) with 2-sided 95% confidence intervals (CIs) were calculated.
A d d i t i o n a l a n a l y s e s .To provide additional ana-lyses regarding the effects of paclitaxel, survival in patients treated with DCB versus PTA was evaluated. As patients treated with PTA were only included in the IN.PACT SFA and IN.PACT SFA Japan trials, these baseline imbalances were adjusted by creating a subgroup that included only patients who met the criteria for pivotal studies (IN.PACT SFA, IN.PACT SFA Japan, and IN.PACT China). The group of patients treated with DCB, termed the standard DCB group, included patients with RCC 2 to 4, with a single, de
novo lesion#20 cm, calcium levels of none to mild,
and no in-stent restenosis. This resulted in a group of
712 standard DCB patients, first reported at the
Vascular Interventional Advances (VIVA) conference
in 2018(13). Survival was compared in the standard
cohort between standard DCB- and PTA-treated pa-tients with the use of propensity scores and IPTW adjustment. Details on the propensity score model are
included in theOnline Appendix.
Upon inspecting the data, we found evidence that the traditional proportional hazard assumption might not be tenable across the 5-year follow-up period. To further evaluate the difference in survival between
the 2 groups, a piecewise exponential model wasfit to
the data. Technical details on the model are included
in theOnline Appendix.
Visit compliance was defined as the proportion of scheduled visits attended and was estimated for both treatment groups at each year and overall. Medication use following the index procedure was captured and reported.
RESULTS
BASELINE CHARACTERISTICS. This independent patient-level analysis included 1,980 patients: 1,837 received DCB and 143 received PTA. The mean age of
the DCB cohort was 68.5 9.8 years; 68.2% (1,253 of
1,837) of patients were male. Baseline characteristics
are summarized inOnline Tables 2 to 5. Overall,
pa-tients treated with a DCB were more likely to have critical limb ischemia (CLI) compared with PTA. DCB subjects were less likely to have hyperlipidemia, coronary artery disease, and diabetes mellitus than those treated with uncoated PTA.
In the overall cohort of 1,837 DCB patients and 143 PTA patients, 181 patients treated with DCB died, compared with 12 patients treated with PTA through the follow-up period up to 5 years. Baseline charac-teristics for those who died were similar between
treatment groups (Online Table 6). PTA patients who
died were more likely to be active smokers than DCB patients who died. DCB patients were more likely to have calcified lesions and diabetes mellitus. There were 32 patients with CLI in the DCB group who died.
Among DCB patients who survived (n¼ 1,656) or
died (n ¼ 181), those in the mortality group were
older; had statistically significantly higher rates of comorbidities, including diabetes mellitus (including insulin dependence), carotid artery disease, coronary
heart disease, renal insufficiency, and
below-the-knee disease; and had a higher incidence of critical limb ischemia (Table 1).
SAFETY OUTCOMES. Safety outcomes of DCB and PTA patients through longest follow-up are reported in Table 2. After adjusting for the intracluster dependence within study using a frailty model with study as a random effect, the MAE composite (including all-cause mortality, major target limb
amputation, CD-TVR, thrombosis) risk was lower in
the DCB group (43.6% vs 45.4%; p ¼ 0.051). Safety
outcomes of the standard cohort are reported in
Table 3.
Causes of death over the length of the follow-up period for all DCB patients included cardiac (3.5%), malignancy (1.8%), respiratory (0.9%), neurological (1.0%), hepatobiliary (0.3%), gastrointestinal (0.6%), renal (0.1%), infection (0.4%), other (0.6%), and
un-known (1.8%) (Table 4). These specific causes of death
were typical for this patient population and were comparable to what is seen in patients treated with PTA. All patients had their causes of deaths adjudi-cated by an independent CEC and none were deemed
related to the device. A total of 7 deaths were adju-dicated as related to the procedure: 1 death on day 1 due to sudden death, a coronary disease on day 7, a myocardial infarction on day 15, a cardiac arrest on day 28, necrotizing fasciitis on day 85, heart failure on day 723, and acutely after a secondary procedure on day 838.
DCB PATIENT MORTALITY AND PACLITAXEL DOSE
Looking specifically at those patients treated with DCB, the overall mean nominal dosage of paclitaxel received was not different between patients who died
TABLE 1 Baseline Characteristics of All Patients Who Survived Versus Died Treated With the Paclitaxel DCB Death (n¼ 181) (192 Target Limbs) (231 Target Lesions) Survival (n¼ 1,656) (1,760 Target Limbs)
(1,973 Target Lesions) p Value
Patient characteristics Age, yrs 72.7 9.4 (178) 68.0 9.8 (1,649) <0.001 Male 68.5 (124/181) 68.2 (1,129/1,656) 1.000 Hypertension 89.4 (161/180) 83.0 (1,371/1,652) 0.026 Hyperlipidemia 70.2 (125/178) 69.5 (1,120/1,611) 0.932 Diabetes mellitus 55.0 (99/180) 39.7 (656/1,653) <0.001
Insulin-dependent diabetes mellitus 31.7 (57/180) 15.4 (254/1,653) <0.001
Carotid artery disease 31.5 (47/149) 21.3 (309/1,448) 0.007
Coronary heart disease 53.5 (91/170) 41.6 (660/1,585) 0.003
Smoking
Active 22.7 (41/181) 33.9 (561/1,655) 0.002
Renal insufficiency (baseline serum creatinine $1.5 ng/dl) 24.2 (39/161) 8.7 (129/1,483) <0.001 Below-the-knee vascular disease of target leg
(stenotic/occluded)
57.6 (98/170) 45.2 (710/1,571) 0.002 Previous peripheral revascularization 55.2 (100/181) 48.1 (796/1,656) 0.072
Rutherford category <0.001 0 0.0 (0/181) 0.0 (0/1,653) 1 0.0 (0/181) 0.1 (1/1,653) 2 24.9 (45/181) 35.1 (580/1,653) 3 57.5 (104/181) 55.4 (915/1,653) 4 13.8 (25/181) 7.7 (128/1,653) 5 3.9 (7/181) 1.8 (29/1,653) 6 0.0 (0/181) 0.0 (0/1,653)
Target limb ABI/TBI (mm Hg ratio) per patient 0.657 0.256 (157) 0.690 0.219 (1,542) 0.118 Pre-procedure per lesion
Target lesion type per lesion
De novo 80.5 (186/231) 78.3 (1,544/1,973) 0.498
Restenotic, nonstented 6.5 (15/231) 7.0 (139/1,973) 0.891
In-stent restenosis 15.5 (30/194) 18.4 (290/1,579) 0.373
Calcification 70.1 (155/221) 67.0 (1,290/1,924) 0.365
Thrombus 0.9 (2/221) 0.9 (18/1,920) 1.000
Reference vessel diameter, mm 5.14 0.67 (231) 5.19 0.68 (1,973) 0.254 Occluded lesion, 100% stenosis 32.9 (76/231) 35.6 (702/1,973) 0.467 Diameter stenosis, % 88.29 11.42 (231) 89.28 11.87 (1,973) 0.229
Lesion length, cm 11.83 9.29 (231) 11.49 8.87 (1,973) 0.584
Values are mean SD (n) or % (n/N). This data is site-reported. Baseline lesion and demographic characteristics of all patients that survived vs died treated with the paclitaxel drug-coated balloons (DCBs).
compared with patients who survived (12,202.1
m
g vs. 11,368.7m
g; p¼ 0.186) (Table 5). A Kaplan-Meier sur-vival analysis stratified nominal paclitaxel dosage into terciles: a low, a mid-, and upper-dose group. Mean dosages for the 3 groups were 5,019.0, 10,007.5,and 19,978.2
m
g, respectively. There was no signifi-cant difference in mortality between groups,
demonstrating no direct effect of levels of nominal paclitaxel dose exposure at the index procedure and survival status in the DCB patients through 5 years
(p ¼ 0.731) (Central Illustration). Similar outcomes
were observed by terciles in the standard cohort (Online Figure 1).
MULTIVARIABLE ANALYSIS. A multivariable anal-ysis of baseline demographic, lesion, and procedural characteristics was performed in all DCB patients
(n¼ 1,837). Predictors of increased risk for mortality
through 5 years included age, insulin-dependent diabetes mellitus, previous target and nontarget limb amputation, renal insufficiency, dialysis status,
coronary artery disease, and hyperlipidemia (Table 6).
Paclitaxel dose levels were not selected by the model selection process; however, when paclitaxel exposure
was forced into the final model to show potential
impact, it did not predict mortality. Indeed, the upper tercile dosage showed less risk compared with the lower tercile.
MORTALITY IN DCB VERSUS PTA. In the unadjusted
analysis of all patients (DCB n ¼ 1,837 and PTA
n ¼ 143), rates of all-cause mortality using
Kaplan-Meier estimates through 5 years were 15.1% and
11.2%, respectively (p ¼ 0.092) (Table 2). In the
adjusted standard cohort (DCB n ¼ 712 and PTA
n ¼ 143), rates of all-cause mortality using
TABLE 2 Safety Outcomes Using Kaplan-Meier Estimates Through 5 Years DCB (n¼ 1,837) 95% CI of DCB, % PTA (n¼ 143) 95% CI of PTA, % Hazard Ratio (95% CI)* p Value* MAE composite† 43.57 (534) (34.70–53.61) 45.39 (59) (31.87–61.46) 0.73 (0.53–1.00) 0.051 Death (all-cause) 15.12 (181) (9.31–24.02) 11.15 (12) (4.86–24.47) 1.70 (0.92–3.15) 0.092 CD-TVR‡ 32.40 (368) (23.68–43.30) 37.17 (49) (23.81–54.82) 0.61 (0.42–0.87) 0.007 Major target limb amputation 1.35 (12) (0.21–8.47) 0.00 (0) (0.00–0.00) — — Thrombosis 4.85 (79) (1.87–12.30) 3.59 (5) (0.70–17.30) 0.74 (0.25–2.16) 0.584 CD-TLR§ 30.99 (350) (22.53–41.67) 32.56 (43) (19.59–50.94) 0.63 (0.43–0.92) 0.017 Any TVR 33.00 (378) (24.18–43.97) 36.78 (49) (23.44–54.49) 0.63 (0.44–0.89) 0.010 Any TLRk 31.62 (359) (23.00–42.46) 34.50 (46) (21.33–52.58) 0.61 (0.42–0.88) 0.008 Values are Kaplan-Meier estimate % (n) unless otherwise specified. Unadjusted safety outcomes through 5 years of all patients treated with DCB compared with PTA are reported. *Hazard ratios and p values are from frailty model with study as random effect. No other covariate was included.†A composite of death from any cause, CD-TVR, target limb major amputation, and thrombosis.‡Defined as any reintervention within the target vessel due to symptoms or drop of ankle-brachial index (ABI) $20% or >0.15 when compared with post-procedure baseline ABI or toe-brachial index (TBI). §Defined as any reintervention at the target lesion due to symptoms or drop of ABI of $20% or >0.15 when compared with post-procedure baseline ABI/TBI. kIncludes clinically-driven and incidental or duplex-driven TLR.
CD-TLR¼ clinically-driven target lesion revascularization; CD-TVR ¼ clinically-driven target vessel revascularization; MAE ¼ major adverse events; TLR ¼ target lesion revascularization; TVR¼ target vessel revascularization.
TABLE 3 Safety Outcomes of Standard Cohort Patients Using Kaplan-Meier Estimates Through 5 Years Standard Paclitaxel DCB
(n¼ 712) 95% CI of DCB, % PTA
(n¼ 143) 95% CI of PTA, % HR (95% CI)* p Value*
MAE composite† 37.83 (163) (29.08–48.18) 47.65 (59) (34.66–62.62) 0.65 (0.48–0.88) 0.005 Death (all-cause) 13.16 (58) (7.81–21.73) 10.98 (12) (4.94–23.43) 1.52 (0.82–2.82) 0.188 CD-TVR‡ 27.59 (110) (19.32–38.46) 40.38 (49) (27.32–56.75) 0.52 (0.37–0.74) <0.001 Major target limb amputation 0.63 (1) (0.04–9.08) 0.00 (0) (0.00–0.00) — — Thrombosis 2.85 (17) (0.82–9.71) 6.31 (5) (1.96–19.34) 0.43 (0.19–0.97) 0.043 CD-TLR§ 25.25 (98) (17.35–35.88) 36.33 (43) (23.44–53.36) 0.51 (0.36–0.72) <0.001 Any TVR 28.15 (114) (19.76–39.12) 39.94 (49) (26.89–56.38) 0.53 (0.38–0.75) <0.001 Any TLRk 25.94 (102) (17.86–36.76) 38.57 (46) (25.58–55.25) 0.48 (0.34–0.67) <0.001 Values are Kaplan-Meier estimate % (number of events). Adjusted safety outcomes through 5 years of the standard cohort are reported. *Hazard ratio (HR) and p values are from frailty model with study as random effect. Kaplan-Meier estimates, HR, and p value are adjusted by inverse probability of treatment weighting.†A composite of death from any cause, CD-TVR, target limb major amputation, and thrombosis.‡Defined as any reintervention within the target vessel due to symptoms or drop of ankle-brachial index (ABI)$20% or >0.15 when compared to post-procedure baseline ABI or toe-brachial index (TBI). §Defined as any reintervention at the target lesion due to symptoms or drop of ABI of$20% or >0.15 when compared with post-procedure baseline ABI/TBI. kIncludes clinically-driven and incidental or duplex-driven TLR.
Kaplan-Meier estimates throughfive years were 13.2%
and 11.0%, respectively (p¼ 0.188) (Table 3).
Mortal-ity outcomes per study through longest follow-up are reported inTable 7.
For the piecewise exponential model, after
inspecting the data in the standard cohort, we chose a knot at 3 years that divided the follow-up times in 2 intervals: 0 to 3 years and 3 to 5 years (i.e., through 0 to 3 years follow-up and 4 to 5 years follow-up). Of
the (n ¼ 12) events observed in the PTA group, 4
events were observed in the first interval and 8
events were observed in the second interval (Online
Figure 5). After adjusting for data structure using study as random effect and for baseline imbalances using IPTW, we found an increase in hazard of death
with DCB in the first time interval hazard ratio
comparing DCB to PTA (HR1: 3.76 [95% CI: 1.22 to
11.55]; p¼ 0.021), which was significant, and an
in-crease in hazard of death with PTA in the second
time interval HR comparing DCB to PTA (HR2: 0.56
[95% CI: 0.22 to 1.43]; p¼ 0.227), which was not
sig-nificant. A test for equality found that the 2 HRs were significantly different in the 2 intervals (p ¼ 0.011) (Online Table 7).
PRELIMINARY ALTERNATIVE HYPOTHESES.
Addi-tional preliminary analyses were conducted to
generate hypotheses about how other factors
following the index procedure may have contributed to mortality, such as interactions with the health care system and management of comorbidities.
Compliance with scheduled follow-up visits was
assessed (Online Table 8). PTA patients exhibited
significantly higher rates of scheduled visit compli-ance compared with DCB patients (95.3% vs. 88.4%; p < 0.001) especially at the 1-, 2-, and 3-year visits (Figure 1A). Visit compliance among DCB patients was higher for patients who survived versus died (88.8%
vs. 84.4%; p¼ 0.009) (Figure 1B). Use of antiplatelet
medication beyond the protocol-mandated time-points was investigated (Table 8). At 6-, 12-, 24-, and
36-month follow-up, PTA patients exhibited signi
fi-cantly higher rates of dual antiplatelet therapy
(aspirinþ clopidogrel, ticlopidine, cilastazol, or
pra-sugrel) administration compared with DCB patients
(p ¼ 0.001 at 12 and 36 months; p < 0.001 at 6 and
24 months).
TABLE 4 Causes of Mortality
Causes of Death
IN.PACT SFA IN.PACT SFA Japan
IN.PACT China (n¼ 139) IN.PACT Global (n¼ 1,259) All DCB Patients (n¼ 1,649) DCB (n¼ 184) PTA (n¼ 104) DCB (n¼ 67) PTA (n¼ 29) All DCB and PTA Patients (n¼ 1,782) Cardiac-related 3.3 (6)* 1.0 (1) 0.0 (0) 0.0 (0) 0.7 (1) 4.1 (51) 3.5 (58) 3.3 (59) Malignancy-related 2.7 (5) 3.9 (4) 4.5 (3) 0.0 (0) 0.0 (0) 1.7 (22) 1.8 (30) 1.9 (34) Respiratory-related 1.6 (3) 0.0 (0) 1.5 (1) 3.4 (1) 0.0 (0) 0.8 (11) 0.9 (15) 0.9 (16) Neurological-related 2.2 (4) 0.0 (0) 0.0 (0) 0.0 (0) 0.0 (0) 1.0 (13) 1.0 (17) 1.0 (17) Hepatobiliary-related 0.5 (1) 0.0 (0) 0.0 (0) 0.0 (0) 0.0 (0) 0.3 (4) 0.3 (5) 0.3 (5) Gastrointestinal-related 1.6 (3) 1.9 (2) 0.0 (0) 0.0 (0) 0.7 (1) 0.4 (5) 0.6 (9) 0.6 (11) Renal-related 0.5 (1) 0.0 (0) 0.0 (0) 0.0 (0) 0.0 (0) 0.0 (0) 0.1 (1) 0.1 (1) Infection-related 1.1 (2)† 0.0 (0) 0.0 (0) 0.0 (0) 0.0 (0) 0.4 (5) 0.4 (7) 0.4 (7) Other 0.5 (1) 1.9 (2) 0.0 (0) 0.0 (0) 0.0 (0) 0.6 (8) 0.6 (9) 0.6 (11) Unknown 1.6 (3) 1.0 (1) 0.0 (0) 3.4 (1) 1.4 (2) 2.0 (25) 1.8 (30) 1.8 (32)
Values are reported as % (n). Site-reported causes of death were previously reported(4,5). In this report, they were confirmed by CEC narratives, resulting in updated numbers of cardiac- and infection-related deaths in IN.PACT SFA. Results of IN.PACT SFA are tabulated through 5 years, IN.PACT SFA Japan and IN.PACT Global through 3 years, and IN.PACT China through 1 year. All deaths were adjudicated by an independent CEC. Procedure-related death was defined as a complication associated with the initial placement of the device or any necessary secondary interventions. This includes morbidity associated with either anesthesia or surgical procedure. Device-related death was defined as a complication associated with the device design as it relates to placement, efficacy, or durability that may involve the delivery system. Drug-related death was defined as a complication associated with a patient’s physiological response to the paclitaxel drug. *There were 3 site-reported causes of death for DCB subjects reported previously in IN.PACT SFA (infarction of the right cerebral hemisphere in the anterior and medialflow region at day 127, sudden death at day 287, and hemorrhagic stroke at day 788). They were reclassified into neurological-related, unknown, and neurological-related system classes in this report, respectively. †Within the IN.PACT SFA study, there were only 3 cases of sepsis/infection-related causes of death through 5 years; 2 cases of sepsis are classified as infection (sepsis at day 374 and septic shock at day 756) and 1 case of biliary sepsis was reclassified into hepatobiliary sepsis in this report.
CEC¼ clinical events committee; DCB ¼ drug-coated balloon; PAD ¼ peripheral artery disease; PTA ¼ percutaneous transluminal angioplasty.
TABLE 5 Nominal Doses of Paclitaxel
Death (n¼ 181) Survival (n¼ 1,656) p Value*
n 181 1,655 Mean SD 12,202.06 7,721.66 11,368.72 7,371.19 0.186 Median (Q1, Q3) 10,298.00 (6,340.00, 16,896.00) 8,979.00 (5,809.00, 15,093.00) Min, max (2,553.00, 43,472.00) (1,850.00, 61,949.00)
Nominal doses of paclitaxel (mg) are reported in all DCB patients who died and survived. Nominal paclitaxel dose per balloon was defined according to the product length and diameter as described in the IN.PACT Admiral In-structions for Use. The balloon lengths and diameters received by each patient were captured in procedure re-cords, and the nominal paclitaxel dose per balloon was added together to define the total dose of paclitaxel received per patient in each index procedure. *p value is from Wilcoxon rank sum test.
DISCUSSION
Paclitaxel is a commonly-used cytotoxic agent origi-nally used in an intravenous application as a chemotherapeutic to treat malignancies. The doses for administration range from a dose per body surface
area of 135 to 175 mg/m2in 3- or 24-h infusions. This
results in total dosage of the drug as a range of 236 to
306 mg for a 65-inch, 150-pound adult, or a range of
302 to 392 mg for a 72-inch, 225-pound adult(14). The
number and schedule of these infusions vary
depending on the malignancy and the patient response. The mean level of paclitaxel received by patients from the DCB in this study by tercile was approximately 5.0 mg in the lower, 10.0 mg in the middle, and 20.0 mg in the upper tercile.
CENTRAL ILLUSTRATION Kaplan-Meier Freedom From All-Cause Death by Paclitaxel Dose in
All DCB Patients
95%
90%
85%
100%
80%
10%
0%
0
36
48
Time After Index Procedure (Months)
Freedom from All
-Cause Death
24
12
60
Number at risk
696
634
526
485
90
49
526
468
407
373
62
29
614
558
501
440
8
3
Adjusted p-value* = 0.731*p-value was from frailty model with study as random effect and IPTW adjusted
88.2%
85.8%
84.2%
DCB Lower Tercile
DCB Mid Tercile
DCB Upper Tercile
Paclitaxel Dose
DCB Lower Tercile
DCB Mid Tercile
DCB Upper Tercile
Distribution of Paclitaxel Dose in Each Paclitaxel Tercile in DCB
N
696
526
614
Mean
µg
5,019.0
10,007.5
19,978.2
Std
µg
1,508.6
1,757.7
6,122.1
Median
µg
4,752.0
9,504.0
18,654.0
Q1, Q3
µg
3,653, 6,924
8,448, 11,618
15,399, 22,705
Range
µg
1,850, 6,951
6,989, 13,822
13,902, 61,949
Schneider, P.A. et al. J Am Coll Cardiol. 2019;73(20):2550–63.
Kaplan-Meier estimates of survival are stratified by nominal paclitaxel dosage terciles with all patient-level DCB data available through 5 years. DCB¼ drug-coated balloon; IPTW ¼ inverse probability of treatment weighting.
The dose-limiting form of toxicity experienced in patients treated with paclitaxel intravenously for
cancer is a low absolute neutrophil count (14). The
other adverse reactions in a pooled analysis of pa-tients with solid tumors receiving paclitaxel were primarily bone marrow–related, although peripheral neuropathy and hypersensitivity reactions were both
found (14). Notably, these reactions can also be
ascribed to the most common excipient for intrave-nous paclitaxel, Cremophor EL, a castor oil deriva-tive, and they occurred in the acute and short-term time frames(15).
Paclitaxel is also used on drug-coated devices in
the coronary arteries, andfirst-generation
paclitaxel-coated stents are no longer used in coronary arteries due to poor outcomes, including the occurrence of
late stent thrombosis(16). It was hypothesized that
the paclitaxel-polymer compound was too effective at reducing restenosis and interfered with the critical
formation of tissue to cover the stent struts (16).
Second-generation coronary stents using paclitaxel analogs are now in use.
The use of paclitaxel as the active component of DCBs for use in the femoropopliteal segment was prompted by the antirestenotic capabilities combined with its hydrophobic, lipophilic qualities. Hydropho-bic medication is combined with a hydrophilic excipient, which allows delivery from the balloon surface to the artery surface, and its lipophilic prop-erty permits uptake into the artery wall. Theoretical safety concerns in the lower extremity vasculature mainly involve distal washout of paclitaxel crystals into the microvasculature, potentially impeding the healing process, especially in patients with ischemic leg ulcers. In the IN.PACT DEEP trial, a trend toward a higher rate of amputations at 1 year was reported in a CLI population whose below-the-knee lesions were treated by IN.PACT Amphirion, a balloon with
different material and coating methodology
compared with IN.PACT Admiral(17). In both the PTA
and DCB arms, amputation-free survival in this CLI
population was >80% at 1 year. The authors
high-lighted the many other factors that contribute to wound healing and eventual amputation in CLI
TABLE 6 Multivariable Analysis for Death in All Patients Treated With DCB Through 5 Years
Predictors of Death Through 5 Years Coefficient Standard Error Hazard Ratio (95% CI) p Value
Age, yrs 0.055 0.009 1.056 (1.038–1.074) <0.001
Insulin-dependent diabetes mellitus (Y vs. N) 0.608 0.175 1.836 (1.304–2.585) <0.001 Previous target limb amputation (Y vs. N) 1.185 0.349 3.270 (1.649–6.485) <0.001 Renal insufficiency (baseline serum creatinine $1.5 ng/dl) (Y vs. N) 0.610 0.212 1.840 (1.216–2.786) 0.004 Previous nontarget limb amputation (per limb) (Y vs. N) 0.958 0.340 2.607 (1.339–5.077) 0.005
On dialysis (Y vs. N) 0.703 0.321 2.019 (1.077–3.787) 0.029
Coronary heart disease (Y vs. N) 0.281 0.158 1.324 (0.971–1.805) 0.076
Hyperlipidemia (Y vs. N) 0.280 0.169 0.756 (0.542–1.054) 0.099
Paclitaxel dose tercile (upper vs. lower)* 0.043 0.180 1.044 (0.734–1.484) 0.812 Paclitaxel dose tercile (mid vs. lower)* 0.013 0.191 0.987 (0.678–1.437) 0.947 Predictors of death through 5 years in all patients treated with DCB are reported. Missing univariate/multivariate predictors were imputed this way: simple imputation was applied for missing values, stratified by study, sex, and treatment arm, using the mean for continuous variables or most common values for dichotomous or categorical variables of the nonmissing values. Multivariate predictors started from the exact same set of variables used to derive IPTW between paclitaxel dose tercile, and were chosen by eliminating 1 of the covariates with the largest p value at a time, until all covariates left in the model except for paclitaxel dose tercile had a p value<0.2. Frailty Cox model with study as random effect was conducted to calculate the hazard ratio and p value. *Paclitaxel dose tercile was forced into multivariate model.
Abbreviations as inTable 4.
TABLE 7 Mortality Rates Through 5 Years by Study
IN.PACT SFA IN.PACT SFA Japan
IN.PACT China IN.PACT Global
DCB PTA p Value DCB PTA p Value
1 yr 1.9 (4/207) 0 0.926 0 0 — 2.9 (4/139) 3.5 (46/1,311)
2 yrs 8.1 (16/198) 0.9 (1/106) 0.008 6.1*(4/66) 3.4 (1/29) 1.000 Study went through 1 yr 7.6 (97/1,275) 3 yrs 10.7 (21/197) 1.9 (2/103) 0.006 6.0*(4/67) 6.9 (2/29) 1.000 Study went through 1 yr 11.4 (144/1,259) 4 yrs 13.0 (24/184) 6.8 (7/103) 0.116 Study went through 3 yrs Study went through 1 yr Complete data not yet available 5 yrs 15.8 (29/184) 9.6 (10/104) 0.156 Study went through 3 yrs Study went through 1 yr Complete data not yet available Values are % (n/N). Mortality rates through 5 years are reported by individual study for all patients. *At 2-year follow-up in IN.PACT SFA Japan, there were 66 evaluable patients. At 3-year follow-up there were 67 evaluable patients. There were no deaths between years 2 and 3.
patients, including pedal circulation status, the loca-tion of the wound or infecloca-tion, and the quality of the wound care program. This concern is challenging to thoroughly investigate mechanistically in humans, although attempts have been made in animal models
to examine particulate embolization (18). Clinical
sequelae have not been seen in these experiments;
however, healthy swine are not identical to human patients with PAD.
Given the known cytotoxicity of paclitaxel, any safety concern must be aggressively and urgently evaluated. We aimed to investigate a potential cor-relation between angioplasty of femoropopliteal ar-teries with a paclitaxel-coated balloon and increased
FIGURE 1 Follow-Up Visit Compliance Through 5 Years Overall and in DCB Patients Who Survived or Died
100% p < 0.001 p < 0.001 p < 0.001 p = 0.264 p = 0.356 p < 0.001 95% 93.2% 97.9% 92.1% 96.8% 91.7% 96.3% 95.3% 96.9% 96.2% 97.4% 88.4% 95.3% 90% 85% 80%
A
DCB (n = 1,837) PTA (n = 143) DCB Death (n = 181) DCB Survival (n = 1,656)Year 1 Year 2 Year 3 Year 4 Year 5 Overall
100% p = 0.131 p = 0.018 p = 0.257 p = 0.433 * p = 0.009 95% 91.0% 93.4% 87.3% 92.4% 86.7% 91.7% 91.7% 95.4% 85.7% 96.2% 84.4% 88.8% 90% 85% 80%
B
Year 1 Year 2 Year 3 Year 4 Year 5 Overall
(A) Follow-up visit compliance rates of drug-coated balloons (DCBs) and percutaneous transluminal angioplasty (PTA) of all patient-level data available up to 5 years. (B) Follow-up visit compliance rates of all patients treated with DCBs comparing survival and mortality using all patient-level data available up to 5 years. Results of IN.PACT SFA are tabulated through 5 years, IN.PACT SFA Japan and IN.PACT Global through 3 years, and IN.PACT China through 1 year. *Due to limited data at 5 years, a p value could not be calculated.
all-cause mortality, as was recently reported (1). In this independent patient-level analysis, there was no significant difference in up to 5-year mortality rates between patients treated with DCB and PTA. Among individual DCB patients, the dose of paclitaxel received during the index procedure did not correlate with mortality risk. All deaths in both treatment groups were independently adjudicated and were not
clustered to a single cause of death. A multivariable analysis demonstrated several significant predictors of mortality. Paclitaxel exposure or dose were not predictors of mortality. Differences in post-index procedure care may have played a role in mortality and urgently requires further investigation.
Further explanation is warranted as to the differing conclusions from the present analysis and the
TABLE 8 Antiplatelet Regimens Through 36 Months
IN.PACT DCB (n¼ 1,837)
PTA
(n¼ 143) Difference (95% CI) p Value*
Discharge ASA 96.4 (1,766/1,832) 98.6 (141/143) 2.2 (10.7 to 6.3) 0.231 Clopidogrel 93.6 (1,715/1,832) 95.8 (137/143) 2.2 (10.7 to 6.3) 0.370 Cilastazol 4.6 (79/1,711) 3.6 (3/83) 1.0 (10.0 to 12.0) 1.000 Prasugrel 0.5 (8/1,621) 0.0 (0/111) 0.5 (9.1 to 10.1) 1.000 Ticlopidine 1.3 (24/1,832) 3.5 (5/143) 2.2 (10.7 to 6.3) 0.054
ASAþ clopidogrel, ticlopidine, cilastazol, or prasugrel 92.6 (1,697/1,832) 97.2 (139/143) 4.6 (13.1 to 3.9) 0.040 30 days ASA 95.2 (1,684/1,768) 97.9 (138/141) 2.6 (11.2 to 6.0) 0.205 Clopidogrel 82.5 (1,458/1,768) 88.7 (125/141) 6.2 (14.7 to 2.4) 0.063 Cilastazol 4.2 (69/1,652) 3.7 (3/82) 0.5 (10.6 to 11.6) 1.000 Prasugrel 0.8 (12/1,557) 0.0 (0/109) 0.8 (8.9 to 10.5) 1.000 Ticlopidine 1.3 (23/1,768) 3.5 (5/141) 2.2 (10.8 to 6.3) 0.051
ASAþ clopidogrel, ticlopidine, cilastazol, or prasugrel 81.7 (1,445/1,768) 90.1 (127/141) 8.3 (16.9 to 0.2) 0.011 6 months ASA 90.5 (1,479/1,634) 99.3 (139/140) 8.8 (17.4 to 0.1) <0.001 Clopidogrel 52.1 (852/1,634) 68.6 (96/140) 16.4 (25.0 to 7.8) <0.001 Cilastazol 5.1 (78/1,529) 3.7 (3/82) 1.4 (9.7 to 12.6) 0.795 Prasugrel 0.6 (8/1,423) 1.9 (2/108) 1.3 (11.1 to 8.5) 0.153 Ticlopidine 1.3 (21/1,634) 3.6 (5/140) 2.3 (10.9 to 6.4) 0.049
ASAþ clopidogrel, ticlopidine, cilastazol, or prasugrel 49.4 (807/1,634) 72.9 (102/140) 23.5 (32.0 to 14.9) <0.001 12 months ASA 88.9 (1,403/1,578) 94.9 (129/136) 5.9 (14.7 to 2.8) 0.029 Clopidogrel 46.9 (740/1,578) 55.9 (76/136) 9.0 (17.7 to 0.2) 0.049 Cilastazol 5.4 (79/1,469) 3.7 (3/81) 1.7 (9.5 to 12.9) 0.797 Prasugrel 0.5 (7/1,367) 1.9 (2/104) 1.4 (11.4 to 8.6) 0.129 Ticlopidine 0.8 (13/1,578) 3.7 (5/136) 2.9 (11.6 to 5.9) 0.011
ASAþ clopidogrel, ticlopidine, cilastazol, or prasugrel 42.8 (676/1,578) 57.4 (78/136) 14.5 (23.2 to 5.8) 0.001 24 months ASA 86.6 (1,090/1,258) 93.7 (118/126) 7.0 (16.2 to 2.1) 0.024 Clopidogrel 35.6 (448/1,258) 54.8 (69/126) 19.1 (28.2 to 10.0) <0.001 Cilastazol 5.9 (69/1,167) 4.1 (3/73) 1.8 (10.0 to 13.6) 0.795 Prasugrel 0.7 (8/1,190) 2.1 (2/94) 1.5 (11.9 to 9.0) 0.163 Ticlopidine 0.9 (11/1,258) 2.4 (3/126) 1.5 (10.6 to 7.6) 0.128
ASAþ clopidogrel, ticlopidine, cilastazol, or prasugrel 30.9 (389/1,258) 54.0 (68/126) 23.0 (32.0 to 13.9) <0.001 36 months ASA 85.4 (988/1,157) 87.1 (108/124) 1.7 (11.0 to 7.6) 0.688 Clopidogrel 34.8 (403/1,157) 48.4 (60/124) 13.6 (22.7 to 4.3) 0.004 Cilastazol 6.9 (74/1,070) 5.6 (4/72) 1.4 (10.6 to 13.3) 0.812 Prasugrel 0.6 (7/1,089) 2.2 (2/92) 1.5 (12.2 to 9.1) 0.151 Ticlopidine 0.6 (7/1,157) 2.4 (3/124) 1.8 (11.1 to 7.4) 0.064
ASAþ clopidogrel, ticlopidine, cilastazol, or prasugrel 30.4 (352/1,157) 45.2 (56/124) 14.7 (23.9 to 5.5) 0.001 Values are % (n/N). Antiplatelet regimens are compared between all DCB and PTA patients through 36 months. Note: Cilastazol was not collected in IN.PACT SFA Phase II. Prasugrel was not collected in IN.PACT China and IN.PACT SFA Japan. Dual antiplatelet therapy (DAPT) was required immediately post-procedure and during follow-up. Nonstented patients required DAPT therapy for a minimum of 1 month and stented patients for a minimum of 3 months. *Fisher exact test is used for binary variables.
recently published meta-analysis that ascribes
increased mortality to paclitaxel (1). The
summary-level meta-analysis of published and presented data from heterogeneous studies of a variety of devices faced several disadvantages not experienced in the present study. Quality and completeness of the input data and the methods of analyses varied, and data heterogeneity was substantial. Assumptions that must be made in a summary-level meta-analysis include but are not limited to the following; missing data, censored patients, unreported values, study normality (whether studies are similar enough to each other to be compared), and estimations of paclitaxel exposure rather than actual calculations of
individual patient doses (Online Appendix). Potential
confounders are the inclusion of DCB and DES in the same study (given the different mechanisms of drug delivery and drug dosing), inclusion of studies initi-ated at disparate timepoints (given steady advances in medical management of PAD patients), and the
paucity of published long-term outcomes with
paclitaxel-based devices.
Only 3 studies of paclitaxel-containing devices have reported summary-level data extended to 5 years: 2 DCB studies (IN.PACT SFA, THUNDER [Local Taxan With Short Time Contact for Reduction of Restenosis in Distal Arteries]), and 1 DES study (Zilver PTX)(7,19,20). Each of these trials was powered for patency at 1 year and not for mortality alone at 5 years, and a subsequent evaluation of mortality is useful but not conclusive. Patient-level data from the IN.PACT SFA trial is included in the present study.
The mortality in the PTA cohort in the IN.PACT SFA trial was the lowest ever reported in a vascular device trial through 3 years (1.9%; 2 of 103), and this stands in contrast to a more expected level of mortality in the experimental group (10.7%; 21 of 197). The imbalance between the number of events in the PTA arm meant that the traditional proportional hazard assumption through 5 years was potentially untena-ble. A piecewise exponential model favored PTA
through 3 years and DCB in years 4 and 5. Thisfinding
was expected given the already reported low rate of mortality in the PTA arm of IN.PACT SFA at 2 and 3
years (4,5). Overall, in the adjusted comparison of
DCB- and PTA-treated patients with similar
charac-teristics, there was no significant difference in
mor-tality through 5 years.
The THUNDER trial, initially reported in 2008, was designed to study patients through 2 years. Through 5-year follow up, 48% of the PTA group was lost to
follow-up(19). The Zilver PTX study included a
dou-ble randomization in which those patients with PTA failure were randomized into 2 further groups: BMS
and DES(20). Those secondarily randomized into DES
were included in the paclitaxel counts in the recent meta-analysis, despite being in a subpopulation where PTA initially failed, arguably a different
sub-population (20). Within this subpopulation of PTA
failures there was no significant difference in mor-tality between the 2 secondary randomization arms at 5 years. Finally, in looking at the Zilver PTX study, DES are coated with paclitaxel differently, and the drug release from a permanent implant is different from the way the drug is deposited on the vessel wall by a DCB.
Although comparative long-term mortality rates in vascular device trials are lacking, there are several adjudicated studies that have reported at 2 and 3 years (Online Figures 2 and 3). Mortality at 2 years in DCB, BMS, DES, and PTA trials ranges from 0.9% (in the PTA arm of IN.PACT SFA) to 15%. At 2 years, the DCB arms of the IN.PACT SFA, Japan, and Global were 8.1%, 6.1%, and 7.6%, respectively, well within the range of mortality seen in other trials for
nonpaclitaxel-containing devices (4,11,21). Mortality
among multiple trials at 3 years ranges from 1.9% (in the PTA arm of IN.PACT SFA) to 14.0%, whereas the mortality in DCB patients in IN.PACT SFA, Japan, and Global studies was 10.7%, 6.0%, and 11.4%, respectively, also within the range of other vascular
device study cohorts (5,22,23). An alternate method
of considering mortality rates in endovascular device
trials is comparing RCTs of BMS and PTA (Online
Figure 4). None of these devices are coated in paclitaxel; yet, an increase in mortality rate is demonstrated for BMS versus PTA across multiple trials at 1 and 2 years. In addition, all of these mortality rates are within the ranges estimated by the recent meta-analysis for paclitaxel-coated de-vices(1).
Another potential confounder in this meta-analysis is the underlying disease itself. PAD is a particularly challenging disease to characterize given the high burden of comorbidities experienced by patients and the relationship between these comorbidities and disease progression as well as mortality. The list of factors affecting mortality rate is long and includes (but is not limited to) the development of new phar-maceuticals or treatment strategies and the updating of guideline recommendations, compliance of pa-tients to exercise and lifestyle changes, physician and
patient behaviors in specific geographies due to
varying standards of care, and the aging population. A comprehensive review of mortality rates throughout the published data estimates that the mortality rate
through 5 years ranges from 10% to 52% (Online
Within this analysis, we generated several hy-potheses about the effect of patient care following an index procedure. First, as was pointed out in the
recent meta-analysis in Online Figure 2 (1), a key
challenge in all of the drug-coated device trials is performance bias, as physicians and participants could not be completely blinded. It is possible that PTA patients were followed-up more rigorously and more effort was spent treating their comorbidities, thus reducing the mortality rate, a hypothesis that
has recently been investigated(24). This is supported
by the significantly different rates of follow-up compliance between DCB and PTA patients overall and between DCB patients who survived and died. In addition, the significantly higher rate of dual anti-platelet therapy use in the PTA group could affect long-term mortality outcomes. These hypotheses require further research; however, they offer practical and immediately tangible solutions to help our pa-tients: close observation and aggressive management of comorbidities.
We welcome and strongly urge additional studies of already-existing data and interrogation of large databases to provide further information about dis-ease progression and mortality in the PAD patient population.
STUDY LIMITATIONS.All studies included herein were adjudicated by clinical events committees, and endpoint definitions were consistent; yet, pooling data from distinct trials has drawbacks. There are some data that were included in this report that have not yet undergone peer review, sharing the same shortcomings as the recently-published
meta-anal-ysis (1). PTA patients were only included in the 2
randomized trials in a 2:1 ratio; these studies were not powered to detect differences in mortality. DCB and PTA patient and lesion populations were heteroge-neous, and the small numbers of PTA control patients (<10%) may not be representative of PTA patients in general, limiting the strength of this analysis of mortality. In addition, the small number of mortality events limited the power to compare DCB and PTA. For the dose-response analysis, all available data were used to compare DCB and PTA; however, adjustment methods and definition of a subset with similar lesion characteristics was necessary to ensure comparable baseline characteristics in the 2 groups. Evaluation of DCB and PTA in larger datasets will
provide more power. Only patients with RCC 2 to 4 were included in these studies (36 patients with RCC 5 were included as protocol deviations). Data are incomplete regarding the potential for paclitaxel-based treatments to have been administered in sub-sequent years for those who failed initial treatment and required repeat revascularization.
CONCLUSIONS
Results from this independent patient-level meta-analysis show no difference in mortality between DCB and PTA at 5 years and no correlation between vary-ing levels of paclitaxel exposure and mortality. This paclitaxel DCB is a safe and effective therapy to treat the symptoms of RCC 2 to 4 femoropopliteal PAD. Data transparency and additional analyses are needed to better understand how other factors influence
long-term outcomes in this complex patient
population.
ACKNOWLEDGMENTSThe authors thank Eric Fer-nandez, MD, Hong Wang, MD, MPH, and Azah Tabah, PhD, for their technical review of this paper; and thank Bridget Wall, PhD, for providing medical writing support.
ADDRESS FOR CORRESPONDENCE: Dr. Peter A. Schneider, Division of Vascular and Endovascular Surgery, University of California at San Francisco, 400 Parnassus Avenue, Suite A-501, San Francisco,
California 94143. E-mail:peterschneidermd@aol.com.
PERSPECTIVES
COMPETENCY IN MEDICAL KNOWLEDGE:
Paclitaxel DCBs have previously been reported to be safe and effective to treat femoropopliteal PAD. However, a recent summary-level meta-analysis re-ported that paclitaxel-containing devices are tied to a higher mortality rate.
TRANSLATIONAL OUTLOOK:Additional research investigating post-procedure care for patients with PAD is warranted. More research on the progression and mortality rates is needed. Patient-level follow-up is mandatory for each individual device to exclude potential side effects related to paclitaxel.
R E F E R E N C E S
1.Katsanos K, Spiliopoulos S, Kitrou P, Krokidis M, Karnabatidis D. Risk of death following application of paclitaxel-coated balloons and stents in the femoropopliteal artery of the leg: a systematic re-view and meta-analysis of randomized controlled trials. J Am Heart Assoc 2018;7:e011245. 2.Feldman DN, Armstrong EJ, Aronow HD, et al. SCAI consensus guidelines for device selection in femoral-popliteal arterial interventions. Catheter Cardiovasc Interv 2018;92:124–40.
3.Bailey SR, Beckman JA, Dao TD, et al. ACC/AHA/ SCAI/SIR/SVM 2018 appropriate use criteria for peripheral artery intervention: a report of the American College of Cardiology Appropriate Use Criteria Task Force, American Heart Association, Society for Cardiovascular Angiography and In-terventions, Society of Interventional Radiology, and Society for Vascular Medicine. J Am Coll Car-diol 2019;73:214–37.
4.Laird JR, Schneider PA, Tepe G, et al. Durability of treatment effect using a drug-coated balloon for femoropopliteal lesions: 24-month results of IN.PACT SFA. J Am Coll Cardiol 2015;66:2329–38. 5.Schneider PA, Laird JR, Tepe G, et al. Treatment effect of drug-coated balloons is durable to 3 years in the femoropopliteal arteries: long-term results of the IN.PACT SFA Randomized Trial. Circ Cardiovasc Interv 2018;11:e005891. 6. Schneider PA. Drug-coated balloons show su-perior four-year outcomes versus angioplasty: re-sults from the IN.PACT SFA Randomized Trial. Paper presented at: Vascular InterVentional Ad-vances; Las Vegas, NV; 2017.
7. Laird JR. 5-year results from the IN.PACT SFA Randomized Trial. Paper presented at: Vascular InterVentional Advances; Las Vegas, NV; 2018.
8.Tepe G, Laird J, Schneider P, et al. Drug-coated balloon versus standard percutaneous trans-luminal angioplasty for the treatment of superfi-cial femoral and popliteal peripheral artery disease: 12-month results from the IN.PACT SFA randomized trial. Circulation 2015;131:495–502.
9.Iida O, Soga Y, Urasawa K, et al. Drug-coated balloon vs standard percutaneous transluminal angioplasty for the treatment of atherosclerotic lesions in the superficial femoral and proximal popliteal arteries: one-year results of the MDT-2113 SFA Japan Randomized Trial. J Endovasc Ther 2018;25:109–17.
10. Chen Z, Tepe G. Safety and efficacy of the IN.PACT Admiral DEB in a Chinese population: 12-month results of the IN.PACT SFA China Study. Paper presented at: VEITH; New York, NY; 2017.
11.Micari A, Brodmann M, Keirse K, et al. Drug-coated balloon treatment of femoropopliteal lesions for patients with intermittent claudication and ischemic rest pain: 2-year results from the IN.PACT Global Study. J Am Coll Cardiol Intv 2018;11:945–53. 12. IN.PACT Admiral Instructions for Use - Revision 1G. 2018. Available at:https://www.accessdata. fda.gov/cdrh_docs/pdf14/P140010c.pdf. Accessed January 24, 2019.
13. Shishehbor M. Total IN.PACT all subjects pooled 1-year analysis. Paper presented at: Vascular InterVentional Advances; November 5–8, 2018; Las Vegas, NV.
14. TAXOL (paclitaxel) Injection Warnings and Safety Information. Bristol Myers-Squibb, 2011. 15.Binder S. Evolution of taxanes in the treatment of metastatic breast cancer. Clin J Oncol Nurs 2013;17 Suppl:9–14.
16.Otsuka F, Vorpahl M, Nakano M, et al. Pa-thology of second-generation everolimus-eluting stents versus first-generation sirolimus- and paclitaxel-eluting stents in humans. Circulation 2014;129:211–23.
17.Zeller T, Baumgartner I, Scheinert D, et al. Drug-eluting balloon versus standard balloon an-gioplasty for infrapopliteal arterial revasculariza-tion in critical limb ischemia: 12-month results from the IN.PACT DEEP randomized trial. J Am Coll Cardiol 2014;64:1568–76.
18.Torii S, Yahagi K, Mori H, et al. Biologic drug effect and particulate embolization of drug-eluting
stents versus drug-coated balloons in healthy swine femoropopliteal arteries. J Vasc Interv Radiol 2018;29:1041–9.e3.
19.Tepe G, Schnorr B, Albrecht T, et al. Angioplasty of femoral-popliteal arteries with drug-coated balloons: 5-year follow-up of the THUNDER trial. J Am Coll Cardiol Intv 2015;8: 102–8.
20.Dake MD, Ansel GM, Jaff MR, et al. Durable clinical effectiveness with paclitaxel-eluting stents in the femoropopliteal artery: 5-year results of the Zilver PTX randomized trial. Circulation 2016;133: 1472–83.
21.Iida O, Soga Y, Urasawa K, et al. Drug-coated balloon versus uncoated percutaneous trans-luminal angioplasty for the treatment of athero-sclerotic lesions in the superficial femoral and proximal popliteal artery: 2-year results of the MDT-2113 SFA Japan randomized trial. Catheter Cardiovasc Interv 2019;93:664–72.
22. Iida O. 3-year results from the MDT-2113 SFA Japan Trial—DCB vs. standard PTA for the treat-ment of atherosclerotic lesions in the SFA/PPA. Paper presented at: Leipzig Interventional Course; 2019; Leipzig, Germany.
23. Tepe G. Three-year results from the IN.PACT Global Study. Paper presented at: CIRSE; 2018; Lisbon, Portugal.
24.Armstrong EJ, Chen DC, Westin GG, et al. Adherence to guideline-recommended therapy is associated with decreased major adverse cardio-vascular events and major adverse limb events among patients with peripheral arterial disease. J Am Heart Assoc 2014;3:e000697.
KEY WORDS drug-coated balloon, femoropopliteal artery, paclitaxel
APPENDIX For an expanded Methods section as well as supplemental tables andfigures, please see the online version of this paper.
